Authors Split Over Anthropic’s $1.5B Copyright Settlement

A federal judge has approved a landmark $1.5 billion copyright settlement between AI company Anthropic and more than 300,000 authors and publishers, resolving claims that the company illegally used pirated copies of nearly 500,000 books to train its Claude AI chatbot. But while the settlement — the largest in U.S. copyright history — delivers substantial compensation, many authors report mixed feelings about whether true justice has been served.

Context: Fair Use vs. Piracy

The case, Bartz v. Anthropic, revolved around two distinct legal questions. In June 2025, U.S. District Judge William Alsup ruled that training AI models on copyrighted text for the purpose of generating new outputs is “spectacularly” transformative and therefore qualifies as fair use — a landmark victory for the AI industry. However, the court also found that Anthropic had illegally downloaded and stored millions of copyrighted books from pirate sites such as Library Genesis. This piracy was not protected by fair use.

Rather than face a jury trial over the piracy question, Anthropic agreed to the $1.5 billion settlement in September 2025. The settlement does not require the company to admit wrongdoing on the fair use question, and because the case settled, Alsup’s fair use ruling was never reviewed by an appeals court, meaning it is persuasive but not binding precedent.

Settlement Details

Judge Araceli Martínez-Olguín granted final approval on July 20, 2026, approving the largest copyright class-action settlement in U.S. history. Under the terms:

• Per-work payout: Approximately $3,000 per work, with authors splitting proceeds with publishers
• Eligible works: 482,460 works; 440,490 (91.3%) were claimed — an exceptionally high rate compared to the typical 10% in class actions
• Class participation: Over 300,000 authors and publishers; only 350 opted out
• Lawyers’ fees: Reduced from an initial request of $300 million to approximately $101 million (about 7% of the settlement fund)
• Plaintiff service awards: Reduced from $50,000 to $15,000 per lead plaintiff

Anthropic is paying the settlement in installments: $300 million already deposited in escrow, $300 million within five days of approval, $450 million by September 25, 2026, and a final $450 million by September 25, 2027, as detailed by Ars Technica.

Mixed Reactions from Authors

Lead plaintiff Charles Graeber, author of The Good Nurse, expressed both pride and frustration. “I was proud to be part of a group that showed that a ragtag bunch of authors joined later by publishers could actually hold together as a class, face a Goliath like Anthropic and get a meaningful number out of them,” he told NPR. But he also noted the personal cost: “A lot of travel, a lot of discussion about what to do and how to proceed and a lot of jobs passed up. I’m much poorer for this settlement, ironically.”

Andrea Bartz, a New York Times bestselling author and fellow lead plaintiff, said the fair use argument makes little sense when AI-generated texts compete directly with human authors. “The algorithm is being used to essentially try to put us out of a job,” she said, calling for international treaties to govern AI and copyright.

Anthropic deputy general counsel Aparna Sridhar defended the outcome, stating that “training AI on books is fair use under copyright law” and that the company is “looking forward to bringing this matter to a close.” The Authors Guild noted that the 91.3% claims rate demonstrates overwhelming author participation.

Unresolved Legal Landscape

Because the case settled, Alsup’s fair use ruling has no binding force on other courts. Multiple lawsuits against Google, Meta, OpenAI, and Midjourney remain ongoing. Just last week, a group of publishers including Hachette and Cengage filed a new class action against Google over its Gemini training data, as reported by TechCrunch.

International Enforcement Gap

Even as U.S. courts wrestle with these questions, the global dimension remains daunting. Chinese AI companies like DeepSeek use “AI distillation” — feeding their models outputs from U.S.-based AI systems rather than directly training on copyrighted books — making enforcement across borders extremely difficult. Intellectual property lawyer Dale Nelson noted that “the government in China controls in large part what the companies are doing,” making enforcement attempts there “very difficult.”

The Licensing Path Forward

The Authors Guild advocates for licensing as a more sustainable solution. Umair Kazi, the Guild’s director of policy and advocacy, explained that “licensing is a way to make sure that training happens legally” and enables rights holders to restrict how their works appear in AI outputs. Some deals already exist — Perplexity AI has licensing agreements with publishers like the Los Angeles Times and Le Monde — but such arrangements remain rare while most major AI companies remain in litigation.

What’s Next

Payouts to class members are expected to begin in late 2026. Meanwhile, the broader legal questions — whether AI training on copyrighted material constitutes fair use, how to compensate creators, and how to enforce rules across borders — remain far from settled. As Bartz put it, her hope is that this lawsuit is “the first of many steps that will create a more fair environment for creatives in the era of AI.”